Robust Audio Watermarking Scheme Based on Wavelet Transforming Stable Feature

In order to improve the watermark invisibility and robustness simultaneously, a robust audio watermarking scheme with self-sychronization which based on discrete wavelet transforming stable feature was proposed. When audio is under attacks, the absolute average value of low frequency sub-bands of discrete wavelet is a steady feature. Thus taking this absolute average value as flags of watermark boundary and domain of embedding watermark is effective and practical. In this scheme, watermark was embedded through adjusting the absolute average value of wavelet transforming low frequency sub-bands in every three audio frames, each of which consists of N audio samples. Similarly watermark is attracted by estimating the relation among the absolute average value of wavelet transforming low frequency sub-bands in every three audio frames. Also two audio frames¿ absolute value of wavelet transforming need to be altered to keep accordance with the global average absolute value of the whole audio, which is adopted to embed flags of start and ending. Based on experimental results, the robust watermark was detected with high accuracy.